Satellite SST and CHL data extractions from selected locations

Marine Biodiversity Observation Network Pole to Pole of the Americas (MBON Pole to Pole)

Written by E. Montes (emontesh@usf.edu) and Eduardo Klein (eklein@usb.ve) on Auguts 28, 2020.

This code pulls data from NOAA’s ERDDAP servers and creates time series plots of sea surface temperature (SST) and chlorophyll-a concentration (CHL), and maps showing the latest available data for the selected region.

Step 1

First, let’s load required libraries

library(readr)
library(rerddap)
library(lubridate)
library(dplyr)
library(flexdashboard)
library(reshape2)
library(leaflet)
library(ggplot2)
library(vegan)
library(xts)
library(dygraphs)
library(plotly)
library(mapdata)

library(RColorBrewer)
palette(brewer.pal(8, "Set2"))

Step 2

Query SST data from ERDDAP

## remove all spaces from string
NoSpaces = function(x){
  return(gsub(" ", "", x))
}

## set site coordinates and time for SST extraction
SSTSiteName = "Patagonia"   ## for the resulting file name
SSTcoords.lon = -64.
SSTcoords.lat = -41.7

SSTstartDate = "2002-06-01"

## set climatological date start-end
SSTclimStartDate = "2002-06-01"
SSTclimEndDate = "2012-12-31"

## set dataset source
SSTsource = info("jplMURSST41")

##
## Get sst 
SST <- griddap(SSTsource, 
              time=c(SSTstartDate, "last"),
              longitude = c(SSTcoords.lon,SSTcoords.lon),
              latitude = c(SSTcoords.lat,SSTcoords.lat),
              fields = "analysed_sst",
              fmt = "csv")

SST = SST[,c(1,4)]
names(SST) = c("time", "SST")

## convert time to a Data object
SST$time = as.Date(ymd_hms(SST$time))

Step 3

Calculate SST climatology

SST.clim = SST %>% filter(time>=ymd(SSTclimStartDate), time<=SSTclimEndDate) %>% 
  group_by(yDay = yday(time)) %>% 
  summarise(SST.mean = mean(SST),
            SST.median = median(SST),
            SST.sd = sd(SST),
            SST.q5 = quantile(SST, 0.05),
            SST.q10 = quantile(SST, 0.10),
            SST.q25 = quantile(SST, 0.25),
            SST.q75 = quantile(SST, 0.75),
            SST.q90 = quantile(SST, 0.90),
            SST.q95 = quantile(SST, 0.95),
            SST.min = min(SST),
            SST.max = max(SST))

Step 4

Plot SST time series

SST.xts = as.xts(SST$SST, SST$time)
dygraph(SST.xts, 
        ylab = "Sea Surface Temperature (Deg C)") %>% 
  dySeries("V1", label ="SST (Deg C)", color = "steelblue") %>%
  dyHighlight(highlightCircleSize = 5, 
              highlightSeriesBackgroundAlpha = 0.2,
              hideOnMouseOut = FALSE) %>% 
  dyOptions(fillGraph = FALSE, fillAlpha = 0.4) %>% 
  dyRangeSelector(dateWindow = c(max(SST$time) - years(5), max(SST$time)))

## subset SST for last year
SST.lastyear = SST %>% filter(year(time)==max(year(time)))

## make the plot
pp = ggplot(SST.clim, aes(yDay, SST.mean))
pp = pp + geom_line() + geom_smooth(span=0.25, se=FALSE, colour="steelblue") +  
  geom_ribbon(aes(ymin=SST.q25, ymax=SST.q75), fill="steelblue", alpha=0.5) +
  geom_line(data=SST.lastyear, aes(yday(time), SST), colour="red") + 
  ylab("Sea Surface Temperature (Deg C)") + xlab("Day of the Year") + 
  theme_bw(base_size = 9) 
ggplotly(pp) %>% plotly::config(displayModeBar = F) 

`geom_smooth()` using method = 'loess' and formula 'y ~ x'

Step 5

Save SST time series data

write_csv(SST, path = paste0(NoSpaces(SSTSiteName), "_SST.csv"))
write_csv(SST.clim, path = paste0(NoSpaces(SSTSiteName), "_Climatology.csv"))

Step 6

Create a map of the latest SST data

sstInfo <- info('jplMURSST41')
# get latest 3-day composite sst
GHRSST <- griddap(sstInfo, latitude = c(-60., -20.), longitude = c(-90., -47.), time = c('last','last'), fields = 'analysed_sst')

mycolor <- colors$temperature
w <- map_data("worldHires", ylim = c(-60., -20.), xlim = c(-90., -47.))
ggplot(data = GHRSST$data, aes(x = lon, y = lat, fill = analysed_sst)) + 
  geom_polygon(data = w, aes(x = long, y = lat, group = group), fill = "grey80") +
  geom_raster(interpolate = FALSE) +
  scale_fill_gradientn(colours = mycolor, na.value = NA) +
  theme_bw() + ylab("latitude") + xlab("longitude") +
  coord_fixed(1.3, xlim = c(-90., -47.),  ylim = c(-60., -20.)) + ggtitle("Latest daily SST data")

Step 7

Query CHL data from ERDDAP

## remove all spaces from string
NoSpaces = function(x){
  return(gsub(" ", "", x))
}

## set site coordinates and time for CHL extraction
CHLSiteName = "Golfo Nuevo"   ## for the resulting file name
CHLcoords.lon = -74
CHLcoords.lat = 12

CHLstartDate = "2012-01-01"

## set climatological date start-end
CHLclimStartDate = "2012-01-01"
CHLclimEndDate = "2016-12-31"

## set dataset source
CHLsource = info("erdMH1chla8day")

##
## Get CHL 
CHL <- griddap(CHLsource, 
               time=c(CHLstartDate, "last"),
               longitude = c(CHLcoords.lon,CHLcoords.lon),
               latitude = c(CHLcoords.lat,CHLcoords.lat),
               fields = "chlorophyll", fmt = "csv")

CHL = CHL[,c(1,4)]
names(CHL) = c("time", "CHL")
CHL = na.omit(CHL)

## convert time to a Data object
CHL$time = as.Date(ymd_hms(CHL$time))

Step 8

Calculate CHL climatology

CHL.clim = CHL %>% filter(time>=ymd(CHLclimStartDate), time<=CHLclimEndDate) %>% 
  group_by(yDay = yday(time)) %>% 
  summarise(CHL.mean = mean(CHL),
            CHL.median = median(CHL),
            CHL.sd = sd(CHL),
            CHL.q5 = quantile(CHL, 0.05),
            CHL.q10 = quantile(CHL, 0.10),
            CHL.q25 = quantile(CHL, 0.25),
            CHL.q75 = quantile(CHL, 0.75),
            CHL.q90 = quantile(CHL, 0.90),
            CHL.q95 = quantile(CHL, 0.95),
            CHL.min = min(CHL),
            CHL.max = max(CHL))

Step 9

Plot CHL time series

CHL.xts = as.xts(CHL$CHL, CHL$time)
dygraph(CHL.xts, 
        ylab = "Chlorophyll a (mg m-3)") %>% 
  dySeries("V1", label ="CHL", color = "steelblue") %>%
  dyHighlight(highlightCircleSize = 5, 
              highlightSeriesBackgroundAlpha = 0.2,
              hideOnMouseOut = FALSE) %>% 
  dyOptions(fillGraph = FALSE, fillAlpha = 0.4) %>% 
  dyRangeSelector(dateWindow = c(max(CHL$time) - years(5), max(CHL$time)))

### CHL Last year with smoothed Climatology {data-width=250}

## subset CHL for last year
CHL.lastyear = CHL %>% filter(year(time)==max(year(time)))

## make the plot
pp = ggplot(CHL.clim, aes(yDay, CHL.mean))
pp = pp + geom_line() + geom_smooth(span=0.25, se=FALSE, colour="steelblue") +  
  geom_ribbon(aes(ymin=CHL.q25, ymax=CHL.q75), fill="steelblue", alpha=0.5) +
  geom_line(data=CHL.lastyear, aes(yday(time), CHL), colour="red") + 
  ylab("Chlorophyll a (mg m-3)") + xlab("Day of the Year") + 
  theme_bw(base_size = 9) 
ggplotly(pp) %>% plotly::config(displayModeBar = F) 

`geom_smooth()` using method = 'loess' and formula 'y ~ x'

#Step 10 Save CHL time series data

write_csv(CHL, path = paste0(NoSpaces(CHLSiteName), "_CHL.csv"))
write_csv(CHL.clim, path = paste0(NoSpaces(CHLSiteName), "_Climatology.csv"))

#Step 11 Create a map of the latest CHL data

require("rerddap")
require("ggplot2")
require("mapdata")

# get latest Monthly chl (VIIRS)
chlaInfo <- info('nesdisVHNSQchlaMonthly')
viirsCHLA <- griddap(chlaInfo, latitude = c(-20., -60.), longitude = c(-90., -47.), time = c('last','last'), fields = 'chlor_a')

# get latest 8-day chl (MODIS)
chlaInfo_8d <- info('erdMH1chla8day')
MODIS_CHLA_8d <- griddap(chlaInfo_8d, latitude = c(-20., -60.), longitude = c(-90., -47.), time = c('last','last'), fields = 'chlorophyll')

# Map monthly chl (VIIRS)
mycolor <- colors$chlorophyll
w <- map_data("worldHires", ylim = c(-60., -20.), xlim = c(-90., -47.))
ggplot(data = viirsCHLA$data, aes(x = lon, y = lat, fill = log(chlor_a))) + 
  geom_polygon(data = w, aes(x = long, y = lat, group = group), fill = "grey80") +
  geom_raster(interpolate = FALSE) +
  scale_fill_gradientn(colours = mycolor, na.value = NA) +
  theme_bw() + ylab("latitude") + xlab("longitude") +
  coord_fixed(1.3, xlim = c(-90., -47.),  ylim = c(-60., -20.)) + ggtitle("Latest VIIRS Monthly Chla")

# Map 8-day chl (MODIS)
mycolor <- colors$chlorophyll
w <- map_data("worldHires", ylim = c(-60., -20.), xlim = c(-90., -47.))
ggplot(data = MODIS_CHLA_8d$data, aes(x = lon, y = lat, fill = log(chlorophyll))) + 
  geom_polygon(data = w, aes(x = long, y = lat, group = group), fill = "grey80") +
  geom_raster(interpolate = FALSE) +
  scale_fill_gradientn(colours = mycolor, na.value = NA) +
  theme_bw() + ylab("latitude") + xlab("longitude") +
  coord_fixed(1.3, xlim = c(-90., -47.),  ylim = c(-60., -20.)) + ggtitle("Latest MODIS 8-day Chla")